Hello, world - and what I plan to write here

This is the first post on a site I have been meaning to set up for a long time. I am a Software Engineer at Therap BD Ltd and a Computer Science graduate from the University of Dhaka, currently applying for PhD programs starting Fall 2027.

I want to use this space mostly for short, focused notes on three threads I keep coming back to in my research:

1. LLMs for competitive programming - what actually fails, and why chain-of-thought reasoning models like the o-series outperform general-purpose models on hard problems (this is the core finding of our LLM-ProS paper).
2. Agentic multi-model feedback - how separating generation from specialized debugging helps, and what trust calibration across critic models looks like in practice (our follow-up work, A-ProS).
3. Research engineering - the boring-but-load-bearing stuff: scrapers, browser automation for Codeforces, SQLite logging, and how to keep a reproducible pipeline alive across hundreds of model calls.

If any of that is interesting to you - or if you are a PI looking for a PhD student with publications and a strong systems background - I would love to hear from you.